I was not aware of CamlIDL, or even of IDL itself until just now.  The
major difference I can see from a quick look at the docs is that the
IDL system converts ML data structures into C data structures and back
again for every function invocation.  Mine simply allocates a string
of the appropriate size for each structure and leaves it in there.

Dave Berry <dave@kal.com> writes:

> Curiousity impels me to ask: How does David's system compare with CamlIDL?
> Is there a need for an FFI interface at a lower-level than CamlIDL?  I'm
> interested because I've worked on FFIs for ML in the past, and we gradually
> moved more and more towards IDL-based solutions.  So I'd be interested to
> hear what limitations people see with this approach (or with the particular
> CamlIDL implementation, as I hope to use that some day).
> 
> Dave.
> 
> 
> -----Original Message-----
> From: jeanmarc.eber@lexifi.com [mailto:jeanmarc.eber@lexifi.com]
> Sent: Friday, April 20, 2001 10:37
> 
> I think that an *easy* FFI (you call it "out of the box") is one of the
> most important stuff to do for Ocaml. Most of us know that there aren't
> any *impossibilities* (more or less anything *can* be done, but you have
> to use exactly the correct macros etc...), but its not .... yes, say
> "out of the box".
-------------------
To unsubscribe, mail caml-list-request@inria.fr.  Archives: http://caml.inria.fr